Parental Perspectives Regarding the Impact of the COVID-19 Pandemic on Their Children

Abstract

Background:

The COVID-19 pandemic has brought profound changes to the health of families worldwide. Yet, there is limited research regarding its impact on children. The pandemic may exacerbate factors associated with excess weight, which is particularly concerning due to the potential association between excess weight and severity of COVID-19 infection. This study investigates parental perspectives of changes in fruit/vegetable (FV) intake, processed food (PF) intake, outdoor playtime (OP), physical activity (PA) levels, and recreational screen time (RST) among children living in Michigan during the pandemic.

Methods:

The study team and community partners developed and distributed a survey using snowball sampling to reach families living largely in Central and Southeastern Michigan. Nonlinear mixed-effects proportional odds models were used to examine associations between child weight status along with demographic/household factors and changes in five weight-related behaviors.

Results:

Parents (n = 1313; representing 2469 children) reported a decrease in OP, FV, and PA levels, while there was an increase in RST and PF intake among their children. Household income was protective against a decrease in OP, PA, and FV but was associated with increased RST. Children's weight status was associated with decreased FV. Age was negatively associated with OP and PA, and positively associated with RST.

Conclusions:

These findings suggest an adverse influence of the pandemic on weight-related behaviors, particularly among adolescents in families with lower incomes and those with excess weight. Further work is needed to measure any impact on BMI trajectory and to identify interventions to reverse negative effects.

Introduction

The ongoing COVID-19 pandemic has rapidly evolved in ways that are having a profound effect on families around the world.^1,2 There is emerging evidence that the health behaviors of children and adolescents, in particular, may have been altered. Their health has likely been affected by myriad factors, including for example, national and local efforts to mitigate contagion, families' adaptation to other changes wrought by the pandemic, and by the COVID-19 infection itself.^3,4 Importantly, some of these changes can potentially impact lifestyle habits that are likely to lead to increases in weight.

Based on data from the National Survey of Children's Health, the prevalence of childhood obesity in 2019–2020 was 16.2%.⁵ Obesity substantially increases the risk of developing hypertension, dyslipidemia, type 2 diabetes, and even cancer.^6,7 Childhood obesity is also linked with depression, poor self-esteem, and anxiety.^8–10 While the links between these health problems and excess weight are well established, there is increasing evidence indicating that obesity is associated with disproportionately severe COVID-19 infection.¹¹

While the severity of COVID-19 infection may be higher in individuals with obesity, further research is needed on how the pandemic itself may lead to an increase in the prevalence of obesity. It is likely that the pandemic may serve as a stressor and exacerbate risk factors associated with excess weight. Research suggests that adolescents tend to have poorer weight outcomes when they are housebound compared with when they are engaged in school activities.¹²

More specifically, African American and Hispanic children are at an increased risk of weight gain when they are housebound, populations who have been disproportionately impacted by the pandemic.^13,14 In addition, it is reasonable to anticipate that during lockdown periods, intake of calorie-dense and processed foods (PFs) may increase, which would further increase the risk of weight gain.¹⁵ Recent research conducted in Italian children and adolescents (n = 41) suggests that lockdown periods adversely impacted specific weight-related behaviors; however, this study mainly focused on children with obesity.¹²

Due to nationwide school closures initiated in March 2020, along with local efforts to blunt the spread of COVID-19, children's access to physical education and team sports was temporarily suspended.^16,17 The shift to online school may also cause a variety of detrimental effects such as depressive symptoms and increased adiposity that are often associated with high levels of screen time.¹⁸

In addition, access to the National School Lunch Program, which serves as the primary food source for >30 million US children and adolescents, was restricted.¹⁹ Given these events, there is strong reason to believe that in the United States, the pandemic may have affected the physical activity (PA), food intake, and recreational screen time (RST) of children, all of which serve as factors that can contribute to excess weight. This study aims to evaluate the perceived impact of the pandemic on these factors known to drive weight gain in children and adolescents, regardless of their current weight status.

Methods

Survey Development

As an initial step to investigate these potential behavioral changes and inform interventions to address them, the study team, in conjunction with community partners in the state of Michigan, developed and distributed a survey to better understand the impact of the pandemic on youth. A 40-item online survey was developed in collaboration with faculty and community partners associated with the Michigan Institute of Clinical and Health Research (MICHR) at the University of Michigan. This approach helped to better ensure that the questions included in the instrument were relevant and as relatable as possible to the health of the community members responding to the survey.

The survey included closed and opened-ended questions with the goal to explore parental perspectives of changes in their children's behavior. Specifically, respondents were asked to report changes in (1) outdoor playtime (OP); (2) PA; (3) RST; (4) fruit/vegetable (FV) intake; and (5) PF intake for their children. The response options utilized a Likert-type scale [Much Less (1), Somewhat Less (2), No Change (3), Somewhat More (4), Much More (5)].

The survey respondents were asked to report their ages and the ages of their children along with their children's height, weight, birth year, and birth month. This information was used to calculate the children's BMI and BMI percentiles, which were used to assess weight status (BMI >85th to <95th percentile indicated overweight status, BMI >95th percentile indicated obesity status). Finally, respondents were asked to indicate their demographics, which included zip code, race/ethnicity, and household income (Appendix A1).

Sample

The ongoing survey was launched on June 11, 2020, to a sample of parents in Michigan with children aged <17 living in their home. An anonymous survey link was distributed through snowball sampling and targeted social media advertisements. We elected to use snowball sampling to distribute the survey through community partners with the aim of evaluating the communities in which MICHR supports research. This sampling method is similar to that used by other studies on the impact of the COVID-19 pandemic, including the distribution of large-scale surveys initially administered to groups associated with academic institutions.²⁰ This method of survey sampling has also been used to study similar dietary changes of communities in Spain.²¹ Respondents were asked to provide data for each child in their household. This analysis includes data collected from June through September 2020.

Statistical Analysis

Descriptive statistics were provided for parents and children using median and interquartile range for continuous variables and frequencies and percentages for categorical variables. Each of the five ordinal outcomes (i.e., FV, PF, PA, OP, RST) were modeled using nonlinear mixed-effects proportional odds models. A random-effects term for the parent was included to account for the clustering of siblings within a household.

Multiple imputations (25 imputations) by fully conditional specification (based on age, education, number of children, and employment status) were used to handle missing data. Overall p-values for the categorical income variable were estimated using the median -2 log-likelihood test across the 25 imputations.

Backward selection was used for variable selection in final models. The following variables were tested as unadjusted predictors of each outcome: age, overweight, income (<$50K vs. $50K–$100K vs. >$100K), education (≥college vs. <college), number of children in household, and employment impacted by pandemic (any household where a respondent indicated an adult in their household was laid off, filed for unemployment, had hours reduced, work closed, was hospitalized and could not work, or had become a full-time caregiver was coded as having employment impacted).

Any variable significant with p < 0.2 was included in a multivariable model. Variables were iteratively removed in descending order of p-value (i.e., removing nonsignificant term with the highest p-value and refitting the model) until any remaining variables were significant with p < 0.05. Proportional odds model estimates were coded, so that higher odds ratios indicated higher levels of the behavior. Analyses were performed in SAS V9.4 (SAS Institute, Inc., Cary, NC, USA).

This study of the impact of the COVID-19 pandemic was reviewed and deemed exempt by the University of Michigan's Institutional Review Board (HUM00181812).

Results

Survey data representing a total of 1313 parents/guardians and 2469 children were collected between June and September 2020, with all adult respondents providing information for all of the minors living in their households. The self-reported zip codes of all the respondents suggested that they were clustered in communities in Central and Southeastern Michigan, particularly near Ann Arbor and Flint.

The majority of respondents were female (85%), White (87%), with an average age of 40 years. Most respondents had a bachelor's degree or higher (79%), and half of the respondents reported a household income of ≥$100,000 (50%). Over 60% of the households represented in the sample had more than one child, with the average age of children being 9 years (range 1–17). Approximately 29% of children had an estimated BMI percentile of ≥85% putting them in the overweight or obesity category. A detailed breakdown of the demographics for survey respondents can be found in Table 1.

Table 1.

Parent and Child Characteristics and Child-Level Behavioral Changes (n = 1313 Parent Respondents; n = 2469 Children)

Characteristic	Parent distribution	Child distribution
Age, median (interquartile range)	40 (35–47)	9 (5–14)
Gender, n (%)
Female	1113 (85)	227 (9)
Male	178 (14)	213 (9)
Genderqueer	2 (<1)	0 (0)
Transgender	1 (<1)	0 (0)
Other	7 (1)	0 (0)
Unknown	12 (1)	2029 (82)
Race, n (%)
American Indian/Alaska Native	4 (<1)	6 (<1)
Asian	33 (3)	42 (2)
Black	64 (5)	145 (6)
Native Hawaiian/Pacific Islander	0 (0)	2 (<1)
White	1146 (87)	1962 (79)
Other	14 (1)	59 (2)
Multiracial	33 (3)	228 (9)
Unknown	19 (1)	25 (1)
Weight
Overweight	—	404 (16)
Healthy weight	—	1017 (41)
Unknown	—	1048 (42)
Income, n (%)
<$20,000	50 (4)	—
$200,001–$50,000	159 (12)	—
$50,001–$100,000	409 (31)	—
>$100,000	661 (50)	—
Unknown	34 (3)	—
Education
<High school	7 (1)	—
High school/GED	34 (3)	—
Some college	151 (12)	—
Associate	92 (7)	—
Bachelor's	402 (31)	—
Master's	455 (35)	—
Doctoral degree	168 (13)	—
Unknown	4 (<1)	—
No. of children, n (%)
1	518 (39)	—
2	538 (41)	—
3	185 (14)	—
4	50 (4)	—
5	14 (14)	—
6	6 (<1)	—
7	2 (<1)	—
Employment impacted, n (%)
Yes	622 (47)	—
No	691 (53)	—

GED, General Educational Development test.

Fruit and Vegetables

Overall, FV intake was reported to have decreased for 23% of children in the study (Table 2). According to the unadjusted model, a higher percentage of respondents whose children were in the overweight/obesity category reported a decrease in FV intake; whereas higher household incomes and higher education levels were associated with a reported increase in the level of FV intake (Table 7). All of these factors, except for education level, remained significant in the adjusted model (Fig. 1). No significant differences were noted with age, number of children in the household, or employment status.

Table 2.

Changes in Child Behavior for Fruit/Vegetable Intake

In general, how has the COVID-19 pandemic changed the following habits for your child	Response, n (%)
	Much less	Somewhat less	No change	Somewhat more	Much more	Unknown
Fruit and vegetable
Overall	107 (4)	420 (17)	1552 (63)	308 (12)	70 (3)	12 (0)
Age <6	16 (2)	108 (13)	606 (73)	75 (9)	22 (3)	5 (1)
Age 6–12	44 (6)	165 (21)	445 (57)	106 (13)	22 (3)	5 (1)
Age 13–18	47 (6)	147 (17)	501 (59)	127 (15)	26 (3)	2 (0)
Healthy weight	32 (3)	170 (17)	624 (61)	161 (16)	27 (3)	3 (0)
Overweight	25 (6)	85 (21)	234 (58)	53 (13)	7 (2)	0 (0)
Income <$50K	48 (11)	117 (26)	205 (46)	45 (10)	25 (6)	2 (0)
Income $50–$100K	39 (5)	123 (17)	453 (62)	96 (13)	21 (3)	4 (1)
Income >$100K	20 (2)	172 (14)	861 (70)	158 (13)	20 (2)	2 (0)
<College	54 (9)	117 (19)	324 (53)	69 (11)	37 (6)	8 (1)
College or higher	53 (3)	303 (16)	1228 (66)	239 (13)	33 (2)	4 (0)
1 child in household	13 (3)	103 (20)	325 (63)	62 (12)	12 (2)	3 (1)
2 children in household	50 (5)	168 (16)	680 (63)	151 (14)	22 (2)	5 (0)
>2 children in household	44 (5)	149 (17)	547 (63)	95 (11)	36 (4)	4 (0)
Employment impacted	72 (6)	214 (18)	694 (58)	156 (13)	48 (4)	4 (0)
Employment not impacted	35 (3)	206 (16)	858 (67)	152 (12)	22 (2)	8 (1)

Table 7.

Unadjusted Nonlinear Mixed-Effects Proportional Odds Model Results

Unadjusted model	Odds ratio [95% CI]	p
Fruit/vegetable intake
Age (per 1 year)	0.98 [0.95–1.00]	0.09
Overweight vs. healthy weight	0.68 [0.50–0.94]	0.02
Income		<0.001
$50K–$100K	3.02 [1.73–5.28]	<0.001
>$100K	3.70 [2.20–6.21]	<0.001
<$50K	Reference	Reference
College or higher vs. <college	1.58 [1.02–2.45]	0.04
No. of children in household (per child)	1.02 [0.85–1.23]	0.83
Employment impacted: yes vs. no	0.87 [0.60–1.25]	0.44
PF intake
Age (per 1 year)	1.03 [1.00–1.06]	0.09
Overweight vs. healthy weight	1.15 [0.81–1.62]	0.44
Income		0.14
$50K–$100K	0.44 [0.21–0.91]	0.03
>$100K	0.34 [0.18–0.67]	0.002
<$50K	Reference	Reference
College or higher vs. <college	0.73 [0.41–1.31]	0.30
No. of children in household (per child)	1.37 [1.08–1.75]	0.01
Employment impacted: yes vs. no	1.08 [0.70–1.67]	0.74
Physical activity
Age (per 1 year)	0.84 [0.82–0.86]	<0.001
Overweight vs. healthy weight	0.82 [0.61–1.09]	0.17
Income		<0.001
$50K–$100K	1.81 [1.03–3.19]	0.04
>$100K	1.83 [1.08–3.10]	0.03
<$50K	Reference	Reference
College or higher vs. <college	1.52 [0.97–2.36]	0.07
No. of children in household (per child)	1.15 [0.95–1.40]	0.14
Employment impacted: yes vs. no	0.84 [0.58–1.22]	0.36
Outside playtime
Age (per 1 year)	0.86 [0.84–0.89]	<0.001
Overweight vs. healthy weight	0.80 [0.60–1.07]	0.13
Income		<0.001
$50K–$100K	2.60 [1.34–5.04]	0.005
>$100K	2.89 [1.55–5.37]	0.001
<$50K	Reference	Reference
College or higher vs. <college	2.44 [1.45–4.10]	0.001
No. of children in household (per child)	1.32 [1.06–1.65]	0.01
Employment impacted: yes vs. no	0.68 [0.45–1.05]	0.08
RST (not schoolwork)
Age (per 1 year)	1.20 [1.16–1.24]	<0.001
Overweight vs. healthy weight	0.89 [0.62–1.27]	0.51
Income		0.004
$50K–$100K	1.23 [0.67–2.25]	0.51
>$100K	1.69 [0.96–2.99]	0.07
<$50K	Reference	Reference
College or higher vs. <college	1.48 [0.92–2.38]	0.11
No. of children in household (per child)	0.96 [0.79–1.18]	0.73
Employment impacted: yes vs. no	0.92 [0.62–1.37]	0.69

Higher odds ratio indicates higher level of behavior during the pandemic. Each of the five outcomes measured by “In general, how has the COVID-19 pandemic changed the following habits for your child's”: with the following ordinal responses 1 = Much less; 2 = Somewhat less; 3 = No change; 4 = Somewhat more; 5 = Much more. Parents were the respondents but provided separate responses for each child. A random-effects term for parent was included to account for the clustering of siblings within household. Multiple imputations (25 imputations) by fully conditional specification (based on age, education, number of children, and employment status) were used to handle missing data. Backward selection was used for variable selection in final models. The following variables were tested as unadjusted predictors of each outcome: age, overweight, income, education, number of children in household, and employment impacted by pandemic. Any variable significant with p < 0.2 was included in a multivariable model. Variables were iteratively removed in descending order of p-value (i.e., removing nonsignificant term with the highest p-value and refitting the model) until any remaining variables were significant with p < 0.05.

Figure 1.

Adjusted nonlinear mixed-effects proportional odds model results. Higher odds ratio indicates higher level of behavior during the pandemic.

Processed Food

PF intake was reported to have increased among 36% of children in the study (Table 3). Considering income as a categorical variable, the unadjusted model revealed that higher household income was associated with a decrease in reported PF intake (Table 7). In addition, the number of children in the household was positively associated with PF intake. This was the only factor that remained significant in the adjusted model (Fig. 1). No significant differences were observed with age, weight status, education, or employment for PF intake.

Table 3.

Changes in Child Behavior for Processed Food Intake

In general, how has the COVID-19 pandemic changed the following habits for your child	Response, n (%)
	Much less	Somewhat less	No change	Somewhat more	Much more	Unknown
PF intake
Overall	110 (4)	328 (13)	1139 (46)	730 (30)	143 (6)	19 (1)
Age <6	25 (3)	79 (9)	479 (58)	216 (26)	25 (3)	8 (1)
Age 6–12	44 (6)	128 (16)	322 (41)	231 (29)	57 (7)	5 (1)
Age 13–18	41 (5)	121 (14)	338 (40)	283 (33)	61 (7)	6 (1)
Healthy weight	49 (5)	167 (16)	429 (42)	319 (31)	49 (5)	4 (0)
Overweight	24 (6)	70 (17)	140 (35)	137 (34)	32 (8)	1 (0)
Income <$50K	28 (6)	58 (13)	133 (30)	154 (35)	67 (15)	2 (0)
Income $50–$100K	41 (6)	75 (10)	345 (47)	226 (31)	43 (6)	6 (1)
Income >$100K	38 (3)	190 (15)	629 (51)	337 (27)	32 (3)	7 (1)
<College	37 (6)	81 (13)	238 (39)	183 (30)	60 (10)	10 (2)
College or higher	73 (4)	247 (13)	901 (48)	547 (29)	83 (4)	9 (0)
1 child in household	32 (6)	71 (14)	230 (44)	163 (31)	16 (3)	6 (1)
2 children in household	52 (5)	150 (14)	485 (45)	321 (30)	59 (5)	9 (1)
>2 children in household	26 (3)	107 (12)	424 (48)	246 (28)	68 (8)	4 (0)
Employment impacted	62 (5)	164 (14)	519 (44)	345 (29)	90 (8)	8 (1)
Employment not impacted	48 (4)	164 (13)	620 (48)	385 (30)	53 (4)	11 (1)

PF, processed food.

Physical Activity

The analysis of changes in children's PA shows similar patterns to children's FV intake. Over half (53%) of the children represented in this sample were reported to have a decrease in their PA levels (Table 4). Based on both the unadjusted (Table 7) and adjusted (Fig. 1) models, a higher household income was associated with increased PA levels, while older age was associated with decreased PA levels. No significant differences were found regarding weight status, education level, number of children in the household, and employment status.

Table 4.

Changes in Child Behavior for Physical Activity

In general, how has the COVID-19 pandemic changed the following habits for your child	Response, n (%)
	Much less	Somewhat less	No change	Somewhat more	Much more	Unknown
Physical activity
Overall	501 (20)	825 (33)	576 (23)	423 (17)	128 (5)	16 (1)
Age <6	60 (7)	199 (24)	355 (43)	147 (18)	62 (7)	9 (1)
Age 6–12	265 (34)	293 (37)	100 (13)	105 (13)	19 (2)	5 (1)
Age 13–18	176 (21)	333 (39)	121 (14)	171 (20)	47 (6)	2 (0)
Healthy weight	227 (22)	381 (37)	189 (19)	171 (17)	46 (5)	3 (0)
Overweight	116 (29)	150 (37)	60 (15)	59 (15)	15 (4)	4 (1)
Income <$50K	137 (31)	129 (29)	73 (17)	62 (14)	39 (9)	2 (0)
Income $50–$100K	135 (18)	250 (34)	185 (25)	120 (16)	42 (6)	4 (1)
Income >$100K	219 (18)	424 (34)	308 (25)	230 (19)	46 (4)	6 (0)
<College	162 (27)	178 (29)	119 (20)	99 (16)	45 (7)	6 (1)
College or higher	339 (18)	647 (35)	457 (25)	324 (17)	83 (4)	10 (1)
1 child in household	106 (20)	178 (34)	135 (26)	72 (14)	18 (3)	9 (2)
2 children in household	217 (20)	366 (34)	248 (23)	179 (17)	63 (6)	3 (0)
>2 children in household	178 (20)	281 (32)	193 (22)	172 (20)	47 (5)	4 (0)
Employment impacted	267 (22)	389 (33)	256 (22)	203 (17)	68 (6)	5 (0)
Employment not impacted	234 (18)	436 (34)	320 (25)	220 (17)	60 (5)	11 (1)

Outdoor Play

Similarly, a decrease in OP was reported for 42% of the children in the study (Table 5). A higher household income, number of children in the household, and higher education levels were positively associated with increased OP, whereas age was negatively associated with OP according to the unadjusted model (Table 7). The adjusted model only indicated a positive correlation with a higher household income and a negative correlation with age. No significant differences were found according to weight status or employment status.

Table 5.

Changes in Child Behavior for Outside Playtime

In general, how has the COVID-19 pandemic changed the following habits for your child	Response, n (%)
	Much less	Somewhat less	No change	Somewhat more	Much more	Unknown
Outside playtime
Overall	428 (17)	621 (25)	583 (24)	537 (22)	286 (12)	14 (1)
Age <6	73 (9)	151 (18)	263 (32)	215 (26)	125 (15)	5 (1)
Age 6–12	214 (27)	229 (29)	192 (24)	110 (14)	38 (5)	4 (1)
Age 13–18	141 (17)	241 (28)	128 (15)	212 (25)	123 (14)	5 (1)
Healthy weight	180 (18)	276 (27)	229 (23)	211 (21)	117 (12)	4 (0)
Overweight	96 (24)	121 (30)	79 (20)	82 (20)	25 (6)	1 (0)
Income <$50K	127 (29)	107 (24)	61 (14)	85 (19)	59 (13)	3 (1)
Income $50–$100K	113 (15)	191 (26)	189 (26)	152 (21)	89 (12)	2 (0)
Income >$100K	178 (14)	313 (25)	318 (26)	286 (23)	133 (11)	5 (0)
<College	145 (24)	143 (23)	122 (20)	119 (20)	74 (12)	6 (1)
College or higher	283 (15)	478 (26)	461 (25)	418 (22)	212 (11)	8 (0)
1 child in household	101 (19)	135 (26)	132 (25)	97 (19)	48 (9)	5 (1)
2 children in household	197 (18)	260 (24)	253 (24)	234 (22)	127 (12)	5 (0)
>2 children in household	130 (15)	226 (26)	198 (23)	206 (24)	111 (13)	4 (0)
Employment impacted	237 (20)	286 (24)	256 (22)	270 (23)	132 (11)	7 (1)
Employment not impacted	191 (15)	335 (26)	327 (26)	267 (21)	154 (12)	7 (1)

Recreational Screen Time

Finally, there was a reported increase in RST for 78% of the children in the study (Table 6). RST was positively correlated with household income and age according to the unadjusted model (Table 7). Both of these factors remained significant in the adjusted model as well (Fig. 1). No significant differences were observed by employment status, education level, number of children in the household, or weight status.

Table 6.

Changes in Child Behavior for Recreational Screen Time

In general, how has the COVID-19 pandemic changed the following habits for your child	Response, n (%)
	Much less	Somewhat less	No change	Somewhat more	Much more	Unknown
RST (not schoolwork)
Overall	37 (1)	59 (2)	418 (17)	943 (38)	998 (40)	14 (1)
Age <6	10 (1)	16 (2)	257 (31)	329 (40)	210 (25)	10 (1)
Age 6–12	14 (2)	21 (3)	83 (11)	258 (33)	409 (52)	2 (0)
Age 13–18	13 (2)	22 (3)	78 (9)	356 (42)	379 (45)	2 (0)
Healthy weight	11 (1)	27 (3)	89 (9)	405 (40)	485 (48)	0 (0)
Overweight	7 (2)	8 (2)	55 (14)	132 (33)	200 (50)	2 (0)
Income <$50K	21 (5)	24 (5)	75 (17)	126 (29)	194 (44)	2 (0)
Income $50K–$100K	7 (1)	14 (2)	142 (19)	290 (39)	278 (38)	5 (1)
Income >$100K	9 (1)	20 (2)	193 (16)	505 (41)	503 (41)	3 (0)
<College	20 (3)	25 (4)	118 (19)	185 (30)	255 (42)	6 (1)
College or higher	17 (1)	34 (2)	300 (16)	758 (41)	743 (40)	8 (0)
1 child in household	11 (2)	9 (2)	113 (22)	178 (34)	204 (39)	3 (1)
2 children in household	11 (1)	26 (2)	135 (13)	432 (40)	466 (43)	6 (1)
>2 children in household	15 (2)	24 (3)	170 (19)	333 (38)	328 (37)	5 (1)
Employment impacted	18 (2)	36 (3)	191 (16)	464 (39)	475 (40)	4 (0)
Employment not impacted	19 (1)	23 (2)	227 (18)	479 (37)	523 (41)	10 (1)

RST, recreational screen time.

Discussion

This study aimed to understand how the COVID-19 pandemic is impacting the health of children in Central and Southeastern Michigan. Using survey data collected from families in the Summer of 2020, we found that the pandemic had a profound impact on children's activity and diet, but that this impact varied across households in meaningful ways.

OP and PA levels decreased for 42% and 53% of the children, respectively. Previous research conducted during the early stages of the pandemic found similar results, suggesting that children performed less PA, had less OP, and displayed more sedentary behaviors.^17,22 Our study findings indicate that a higher household income was a protective factor against the effects of the pandemic on children's PA and OP.

In addition, other factors such as the number of children in the household and higher parental education levels had favorable effects. Age was negatively associated with PA and OP. The reasons for this inverse relationship are not fully known, but factors such as greater parental influence on the PA and OP of younger children and less dependence on organized team sports for younger children may have led to a smaller decrease in their PA and OP compared with adolescents.

The effects of the pandemic on children's diets were less profound, however still significant. FV intake decreased for 23% of the children, whereas there was no reported change for 63% of the children. PF intake was also reported to increase for 36% of the children represented in the study. Similarly to OP and PA, a higher household income was a protective factor against a decrease in FV intake and against an increase in PF intake.

In contrast, a higher number of kids in the household were associated with increased PF intake, and a weight status of “overweight or obesity” was associated with decreased FV intake. Although a steeper increase in PF intake is often seen in times of confinement, our results indicated that PF intake mostly remained unchanged (46%).²³ A possible explanation for these findings is a lowered consumption of restaurant and fast foods due to business closures during the pandemic.²⁴ In addition, since the sample represented in this study mostly consisted of higher income households, it is likely that these families did not harbor a disproportionate burden of the pandemic and were able to access fresh foods throughout quarantine.

RST was reported to increase for 78% of the children in the study. Previous research analyzing the effects of the pandemic on Polish children found similar results, indicating children spent more time watching movies and playing online games.²⁴ Our study highlights increased levels of RST in children from higher income households. When analyzing age as a categorical factor, the study results showed that a child's age was correlated with increased RST. However, the multivariant analysis demonstrated decreased RST in higher age groups.

The relationships driving the health behaviors of children in households are complex, but the results presented here suggest that parental income and children's age may mitigate the impact that the COVID-19 pandemic is having on the health of children and adolescents. Our findings from this study could help inform resource allocation strategies during the pandemic. To date, most of the allocation guidelines such as the New York State Task Force Ventilator Allocation Guidelines and Interim Pennsylvania Crisis Standard of Care for Pandemic Guidelines, have appropriately focused on issues such as the availability of medications, beds, and ventilators when hospitals have reached maximum capacity and pivoting to address COVID-19 vaccinations.^25–29 However, the impact of the pandemic on youth, particularly among lower income populations, may be long lasting and warrant specific interventions to ameliorate the impact on their health.

Research conducted in Spain found that districts with lower incomes per capita had a higher incidence of COVID-19.³⁰ A similar trend was seen across the United States where morbidity and mortality due to COVID-19 were highest in the Bronx, the borough with the greatest proportion of persons living in poverty along with the greatest proportion of racial/ethnic minorities.³¹ Families with lower household incomes often do not have the space to social distance and contain the spread of infection. This issue is often translated from the home into the community.

Overcrowding in schools in such communities contributes to the spread of infection.³² In addition, attendance at outdoor playgrounds and parks is often limited due to a lack of organized activities, park features, or public intoxication and crime.³³ Food deserts, often seen in low-income communities, and inflation brought on by the pandemic affecting various food items can further lead to food insecurity.

The factors discussed above should be considered when deciding how resources should be allocated during the pandemic. This can be done in the form of neighborhood development by providing police patrolling in parks along with organized activities, cheap and accessible transportation to overcome food insecurity, and stocking community centers with essential resources such as masks, personal protective equipment (PPE), and fresh produce.^34,35 A larger focus on parental education regarding safe hygiene practices and meeting the educational and social needs of their children could also be beneficial.

In addition, strategies that address inequities in general may have an impact on these findings. Although things such as COVID-19 relief money, food-bank resources, and rent assistance programs have been implemented, the impact on health-related choices is not yet known. Future research is needed to understand how changes in children's behavior during the pandemic could lead to excess weight and increased obesity prevalence, particularly among low-income households along with minority and ethnic populations not represented in the study.

Limitations

The results of this study should be considered within the context of certain limitations. Our survey population consisted mainly of White, higher income families in Southeastern Michigan, which limits the generalizability of the study findings. While the sampling methods used for this research enabled the study team to better understand the health behaviors of children living in the communities, they did not lead to a representative sample. Our survey did not include a question about the gender of the respondents or their children. Since the domains investigated in this study could be influenced by gender, further work is needed to see how gender impacts these weight-related behaviors.

The study team did not define PF intake, hence the perception of changes in children's behavior may be subjective based on self-selection of participants. The survey questions aimed to gauge the overall change in children's behaviors, but the wording of the questions did not allow the authors to assess baseline levels. Also, the survey did not inquire about the environment for children's OP and PA, therefore further research can be conducted to assess how their surroundings affect their behaviors.

The findings of this study can be used to inform the models of clinical and translational research to better understand how the COVID-19 pandemic is affecting pediatric and adolescent health. However, due to these limitations, the precise estimates found through this study cannot serve as benchmarks for measuring the pandemic's impact on communities within Central and Southeastern Michigan.

Conclusion

Our findings suggest that the COVID-19 pandemic is affecting the FV intake, PF intake, PA, OP, and RST levels for children in Central and Southeast Michigan. Further research is needed to explore the effects of the pandemic on families with lower household incomes and families of different ethnic and racial backgrounds. The need for longitudinal research that builds on the findings of this study is timely, particularly given the lasting and crippling effects the pandemic is having on the social and educational lives of children and adolescents.

Appendix A1

Question #3—This survey is only for primary caregivers who have children under the age of 18 living in their home. Do you have at least one child <18 years of age living in your home? (Y/N)

Question #4—Please provide the ages of all children (<18 years of age) who live in your household, even if only part-time.

Question #5—Please provide the ages of all adults (18+) living in your household (please list yourself as “self”).

Question #6—Please list the relationship of the following (child) to you:

• - Options included (1) son/daughter, (2) grandchild, (3) other (explain below).

Question #8—How has the COVID-19 pandemic impacted the employment status of the adult members of your household? (Check all that apply)

• - No change

• - Laid off

• - Filed for unemployment

• - Hours were reduced

• - Workplace permanently closed

• - Transitioned to work from home

• - Hospitalized—unable to work

• - Unable to work due to caring for family

• - Other (please specify below)

Question #14—In general, how has the COVID-19 pandemic changed the following habits for your child?

Question #33—What is the zip code of your current residence?

Question #34—How would you best describe the race of everyone in your household? (Check all that apply)

• - Options included Asian, American Indian/Alaska Native, Black/African American, Native Hawaiian/Pacific Islander, White, Other (please explain).

Question #35—Is anyone in your household Hispanic or Latino?

• - Options included Yes, No, Unsure.

Question #36—To calculate the BMI of the children living in your household, please provide the following information below (for each child).

• - Height (ft), Height (inches), Weight (lbs), Month of Birth, Year of Birth

Question #37—What is your gender?

• - Options included Female, Male, Genderqueer, Intersex, Transgender, Other (please specific below).

Question #38—What is the highest level of school you have completed?

• - Less than high school diploma

• - High school diploma or General Education Development (GED) test

• - Some college or technical school, but no degree

• - Associate or technical degree (including LPN, RN, and 1–3 years nursing certificate program)

• - Bachelor's degree (BA, BS, BSN)

• - Master's degree (MA, MS, MENG, MED, MSW, MSN)

• - Doctoral degree (PhD, MD, JD, DMD, DDS, DVM)

Question #39—What category best describes the total annual income from all sources for your household before taxes?

• - <$20,000

• - $20,001–$50,000

• - $50,000–$100,000

• - >$100,000

• - I do not know

Authors' Contributions

S.T. drafted the initial article, made a substantial contribution to the analysis and interpretation of the data, and reviewed and revised the article. J.P.T. and E.S. made a substantial contribution to the analysis and interpretation of the data, and contributed to reviewing and revising the article. A.M. participated in conceptualizing and designing the study, made a substantial contribution to the analysis and interpretation of the data, and also reviewed and revised the article. T.K.-D. participated in conceptualizing and designing the study, and also made a substantial contribution to the analysis and interpretation of the data. D.V. participated in conceptualizing and designing the study, and also reviewed and revised the article. N.G., V.N., K.K., B.C., E.G.-M., L.E., S.B., K.J., A.S., E.H.D., and P.P. participated in conceptualizing and designing the study, and also reviewed and revised the article. S.J.W. made a substantial contribution to the design of the study, drafting of the article, and also provided the final approval. All authors approved the final article as submitted and agreed to be accountable for all aspects of the work.

Funding Information

This project was supported by grant number # UL1TR002240 (Julie Lumeng; George Mashour, PIs) from the National Center for Advancing Translational Sciences (NCATS).

Author Disclosure Statement

No competing financial interests exist.